| 1 |
package Whatpm::ContentChecker; |
| 2 |
use strict; |
| 3 |
our $VERSION=do{my @r=(q$Revision: 1.50 $=~/\d+/g);sprintf "%d."."%02d" x $#r,@r}; |
| 4 |
|
| 5 |
require Whatpm::URIChecker; |
| 6 |
|
| 7 |
## ISSUE: How XML and XML Namespaces conformance can (or cannot) |
| 8 |
## be applied to an in-memory representation (i.e. DOM)? |
| 9 |
|
| 10 |
## TODO: Conformance of an HTML document with non-html root element. |
| 11 |
|
| 12 |
my $HTML_NS = q<http://www.w3.org/1999/xhtml>; |
| 13 |
my $XML_NS = q<http://www.w3.org/XML/1998/namespace>; |
| 14 |
my $XMLNS_NS = q<http://www.w3.org/2000/xmlns/>; |
| 15 |
|
| 16 |
my $Namespace = { |
| 17 |
q<http://www.w3.org/2005/Atom> => {module => 'Whatpm::ContentChecker::Atom'}, |
| 18 |
$HTML_NS => {module => 'Whatpm::ContentChecker::HTML'}, |
| 19 |
$XML_NS => {loaded => 1}, |
| 20 |
$XMLNS_NS => {loaded => 1}, |
| 21 |
}; |
| 22 |
|
| 23 |
our $AttrChecker = { |
| 24 |
$XML_NS => { |
| 25 |
space => sub { |
| 26 |
my ($self, $attr) = @_; |
| 27 |
my $value = $attr->value; |
| 28 |
if ($value eq 'default' or $value eq 'preserve') { |
| 29 |
# |
| 30 |
} else { |
| 31 |
## NOTE: An XML "error" |
| 32 |
$self->{onerror}->(node => $attr, level => 'error', |
| 33 |
type => 'invalid attribute value'); |
| 34 |
} |
| 35 |
}, |
| 36 |
lang => sub { |
| 37 |
my ($self, $attr) = @_; |
| 38 |
my $value = $attr->value; |
| 39 |
if ($value eq '') { |
| 40 |
# |
| 41 |
} else { |
| 42 |
require Whatpm::LangTag; |
| 43 |
Whatpm::LangTag->check_rfc3066_language_tag ($value, sub { |
| 44 |
my %opt = @_; |
| 45 |
my $type = 'LangTag:'.$opt{type}; |
| 46 |
$type .= ':' . $opt{subtag} if defined $opt{subtag}; |
| 47 |
$self->{onerror}->(node => $attr, type => $type, |
| 48 |
value => $opt{value}, level => $opt{level}); |
| 49 |
}); |
| 50 |
} |
| 51 |
|
| 52 |
## NOTE: "The values of the attribute are language identifiers |
| 53 |
## as defined by [IETF RFC 3066], Tags for the Identification |
| 54 |
## of Languages, or its successor; in addition, the empty string |
| 55 |
## may be specified." ("may" in lower case) |
| 56 |
## NOTE: Is an RFC 3066-valid (but RFC 4647-invalid) language tag |
| 57 |
## allowed today? |
| 58 |
|
| 59 |
## TODO: test data |
| 60 |
|
| 61 |
if ($attr->owner_document->manakai_is_html) { # MUST NOT |
| 62 |
$self->{onerror}->(node => $attr, type => 'in HTML:xml:lang'); |
| 63 |
## TODO: Test data... |
| 64 |
} |
| 65 |
}, |
| 66 |
base => sub { |
| 67 |
my ($self, $attr) = @_; |
| 68 |
my $value = $attr->value; |
| 69 |
if ($value =~ /[^\x{0000}-\x{10FFFF}]/) { ## ISSUE: Should we disallow noncharacters? |
| 70 |
$self->{onerror}->(node => $attr, |
| 71 |
type => 'invalid attribute value'); |
| 72 |
} |
| 73 |
## NOTE: Conformance to URI standard is not checked since there is |
| 74 |
## no author requirement on conformance in the XML Base specification. |
| 75 |
}, |
| 76 |
id => sub { |
| 77 |
my ($self, $attr) = @_; |
| 78 |
my $value = $attr->value; |
| 79 |
$value =~ s/[\x09\x0A\x0D\x20]+/ /g; |
| 80 |
$value =~ s/^\x20//; |
| 81 |
$value =~ s/\x20$//; |
| 82 |
## TODO: NCName in XML 1.0 or 1.1 |
| 83 |
## TODO: declared type is ID? |
| 84 |
if ($self->{id}->{$value}) { ## NOTE: An xml:id error |
| 85 |
$self->{onerror}->(node => $attr, level => 'error', |
| 86 |
type => 'duplicate ID'); |
| 87 |
push @{$self->{id}->{$value}}, $attr; |
| 88 |
} else { |
| 89 |
$self->{id}->{$value} = [$attr]; |
| 90 |
} |
| 91 |
}, |
| 92 |
}, |
| 93 |
$XMLNS_NS => { |
| 94 |
'' => sub { |
| 95 |
my ($self, $attr) = @_; |
| 96 |
my $ln = $attr->manakai_local_name; |
| 97 |
my $value = $attr->value; |
| 98 |
if ($value eq $XML_NS and $ln ne 'xml') { |
| 99 |
$self->{onerror} |
| 100 |
->(node => $attr, level => 'NC', |
| 101 |
type => 'Reserved Prefixes and Namespace Names:=xml'); |
| 102 |
} elsif ($value eq $XMLNS_NS) { |
| 103 |
$self->{onerror} |
| 104 |
->(node => $attr, level => 'NC', |
| 105 |
type => 'Reserved Prefixes and Namespace Names:=xmlns'); |
| 106 |
} |
| 107 |
if ($ln eq 'xml' and $value ne $XML_NS) { |
| 108 |
$self->{onerror} |
| 109 |
->(node => $attr, level => 'NC', |
| 110 |
type => 'Reserved Prefixes and Namespace Names:xmlns:xml='); |
| 111 |
} elsif ($ln eq 'xmlns') { |
| 112 |
$self->{onerror} |
| 113 |
->(node => $attr, level => 'NC', |
| 114 |
type => 'Reserved Prefixes and Namespace Names:xmlns:xmlns='); |
| 115 |
} |
| 116 |
## TODO: If XML 1.0 and empty |
| 117 |
}, |
| 118 |
xmlns => sub { |
| 119 |
my ($self, $attr) = @_; |
| 120 |
## TODO: In XML 1.0, URI reference [RFC 3986] or an empty string |
| 121 |
## TODO: In XML 1.1, IRI reference [RFC 3987] or an empty string |
| 122 |
## TODO: relative references are deprecated |
| 123 |
my $value = $attr->value; |
| 124 |
if ($value eq $XML_NS) { |
| 125 |
$self->{onerror} |
| 126 |
->(node => $attr, level => 'NC', |
| 127 |
type => 'Reserved Prefixes and Namespace Names:=xml'); |
| 128 |
} elsif ($value eq $XMLNS_NS) { |
| 129 |
$self->{onerror} |
| 130 |
->(node => $attr, level => 'NC', |
| 131 |
type => 'Reserved Prefixes and Namespace Names:=xmlns'); |
| 132 |
} |
| 133 |
}, |
| 134 |
}, |
| 135 |
}; |
| 136 |
|
| 137 |
## ISSUE: Should we really allow these attributes? |
| 138 |
$AttrChecker->{''}->{'xml:space'} = $AttrChecker->{$XML_NS}->{space}; |
| 139 |
$AttrChecker->{''}->{'xml:lang'} = $AttrChecker->{$XML_NS}->{lang}; |
| 140 |
$AttrChecker->{''}->{'xml:base'} = $AttrChecker->{$XML_NS}->{base}; |
| 141 |
$AttrChecker->{''}->{'xml:id'} = $AttrChecker->{$XML_NS}->{id}; |
| 142 |
|
| 143 |
## ANY |
| 144 |
our $AnyChecker = sub { |
| 145 |
my ($self, $todo) = @_; |
| 146 |
my $el = $todo->{node}; |
| 147 |
my $new_todos = []; |
| 148 |
my @nodes = (@{$el->child_nodes}); |
| 149 |
while (@nodes) { |
| 150 |
my $node = shift @nodes; |
| 151 |
$self->_remove_minuses ($node) and next if ref $node eq 'HASH'; |
| 152 |
|
| 153 |
my $nt = $node->node_type; |
| 154 |
if ($nt == 1) { |
| 155 |
my $node_ns = $node->namespace_uri; |
| 156 |
$node_ns = '' unless defined $node_ns; |
| 157 |
my $node_ln = $node->manakai_local_name; |
| 158 |
if ($self->{minuses}->{$node_ns}->{$node_ln}) { |
| 159 |
$self->{onerror}->(node => $node, type => 'element not allowed'); |
| 160 |
} |
| 161 |
push @$new_todos, {type => 'element', node => $node}; |
| 162 |
} elsif ($nt == 5) { |
| 163 |
unshift @nodes, @{$node->child_nodes}; |
| 164 |
} |
| 165 |
} |
| 166 |
return ($new_todos); |
| 167 |
}; # $AnyChecker |
| 168 |
|
| 169 |
our $ElementDefault = { |
| 170 |
checker => sub { |
| 171 |
my ($self, $todo) = @_; |
| 172 |
$self->{onerror}->(node => $todo->{node}, level => 'unsupported', |
| 173 |
type => 'element'); |
| 174 |
return $AnyChecker->($self, $todo); |
| 175 |
}, |
| 176 |
attrs_checker => sub { |
| 177 |
my ($self, $todo) = @_; |
| 178 |
for my $attr (@{$todo->{node}->attributes}) { |
| 179 |
my $attr_ns = $attr->namespace_uri; |
| 180 |
$attr_ns = '' unless defined $attr_ns; |
| 181 |
my $attr_ln = $attr->manakai_local_name; |
| 182 |
my $checker = $AttrChecker->{$attr_ns}->{$attr_ln} |
| 183 |
|| $AttrChecker->{$attr_ns}->{''}; |
| 184 |
if ($checker) { |
| 185 |
$checker->($self, $attr); |
| 186 |
} else { |
| 187 |
$self->{onerror}->(node => $attr, level => 'unsupported', |
| 188 |
type => 'attribute'); |
| 189 |
} |
| 190 |
} |
| 191 |
}, |
| 192 |
}; |
| 193 |
|
| 194 |
my $HTMLTransparentElements = { |
| 195 |
$HTML_NS => {qw/ins 1 font 1 noscript 1/}, |
| 196 |
## NOTE: |html:noscript| is transparent if scripting is disabled |
| 197 |
## and not in |head|. |
| 198 |
}; |
| 199 |
|
| 200 |
our $Element = {}; |
| 201 |
|
| 202 |
sub check_document ($$$) { |
| 203 |
my ($self, $doc, $onerror) = @_; |
| 204 |
$self = bless {}, $self unless ref $self; |
| 205 |
$self->{onerror} = $onerror; |
| 206 |
|
| 207 |
$self->{must_level} = 'm'; |
| 208 |
$self->{fact_level} = 'f'; |
| 209 |
$self->{should_level} = 's'; |
| 210 |
$self->{good_level} = 'w'; |
| 211 |
|
| 212 |
my $docel = $doc->document_element; |
| 213 |
unless (defined $docel) { |
| 214 |
## ISSUE: Should we check content of Document node? |
| 215 |
$onerror->(node => $doc, type => 'no document element'); |
| 216 |
## ISSUE: Is this non-conforming (to what spec)? Or just a warning? |
| 217 |
return { |
| 218 |
class => {}, |
| 219 |
id => {}, table => [], term => {}, |
| 220 |
}; |
| 221 |
} |
| 222 |
|
| 223 |
## ISSUE: Unexpanded entity references and HTML5 conformance |
| 224 |
|
| 225 |
my $docel_nsuri = $docel->namespace_uri; |
| 226 |
$docel_nsuri = '' unless defined $docel_nsuri; |
| 227 |
unless ($Namespace->{$docel_nsuri}->{loaded}) { |
| 228 |
if ($Namespace->{$docel_nsuri}->{module}) { |
| 229 |
eval qq{ require $Namespace->{$docel_nsuri}->{module} } or die $@; |
| 230 |
} else { |
| 231 |
$Namespace->{$docel_nsuri}->{loaded} = 1; |
| 232 |
} |
| 233 |
} |
| 234 |
my $docel_def = $Element->{$docel_nsuri}->{$docel->manakai_local_name} || |
| 235 |
$Element->{$docel_nsuri}->{''} || |
| 236 |
$ElementDefault; |
| 237 |
if ($docel_def->{is_root}) { |
| 238 |
# |
| 239 |
} elsif ($docel_def->{is_xml_root}) { |
| 240 |
unless ($doc->manakai_is_html) { |
| 241 |
# |
| 242 |
} else { |
| 243 |
$onerror->(node => $docel, type => 'element not allowed:root:xml'); |
| 244 |
} |
| 245 |
} else { |
| 246 |
$onerror->(node => $docel, type => 'element not allowed:root'); |
| 247 |
} |
| 248 |
|
| 249 |
## TODO: Check for other items other than document element |
| 250 |
## (second (errorous) element, text nodes, PI nodes, doctype nodes) |
| 251 |
|
| 252 |
my $return = $self->check_element ($docel, $onerror); |
| 253 |
|
| 254 |
my $charset_name = $doc->input_encoding; |
| 255 |
if (defined $charset_name) { |
| 256 |
require Message::Charset::Info; |
| 257 |
my $charset = $Message::Charset::Info::IANACharset->{$charset_name}; |
| 258 |
|
| 259 |
if ($doc->manakai_is_html and |
| 260 |
not $doc->manakai_has_bom and |
| 261 |
not defined $doc->manakai_charset) { |
| 262 |
unless ($charset->{is_html_ascii_superset}) { |
| 263 |
$onerror->(node => $doc, level => $self->{must_level}, |
| 264 |
type => 'non ascii superset:'.$charset_name); |
| 265 |
} |
| 266 |
|
| 267 |
if (not $self->{has_charset} and |
| 268 |
not $charset->{iana_names}->{'us-ascii'}) { |
| 269 |
$onerror->(node => $doc, level => $self->{must_level}, |
| 270 |
type => 'no character encoding declaration:'.$charset_name); |
| 271 |
} |
| 272 |
} |
| 273 |
|
| 274 |
if ($charset->{iana_names}->{'utf-8'}) { |
| 275 |
# |
| 276 |
} elsif ($charset->{iana_names}->{'jis_x0212-1990'} or |
| 277 |
$charset->{iana_names}->{'x-jis0208'} or |
| 278 |
$charset->{iana_names}->{'utf-32'} or ## ISSUE: UTF-32BE? UTF-32LE? |
| 279 |
$charset->{is_ebcdic_based}) { |
| 280 |
$onerror->(node => $doc, |
| 281 |
type => 'character encoding:'.$charset_name, |
| 282 |
level => $self->{should_level}); |
| 283 |
} elsif ($charset->{iana_names}->{'cesu-8'} or |
| 284 |
$charset->{iana_names}->{'utf-8'} or ## ISSUE: UNICODE-1-1-UTF-7? |
| 285 |
$charset->{iana_names}->{'bocu-1'} or |
| 286 |
$charset->{iana_names}->{'scsu'}) { |
| 287 |
$onerror->(node => $doc, |
| 288 |
type => 'character encoding:'.$charset_name, |
| 289 |
level => $self->{must_level}); |
| 290 |
} else { |
| 291 |
$onerror->(node => $doc, |
| 292 |
type => 'character encoding:'.$charset_name, |
| 293 |
level => $self->{good_level}); |
| 294 |
} |
| 295 |
} |
| 296 |
|
| 297 |
return $return; |
| 298 |
} # check_document |
| 299 |
|
| 300 |
sub check_element ($$$) { |
| 301 |
my ($self, $el, $onerror) = @_; |
| 302 |
$self = bless {}, $self unless ref $self; |
| 303 |
$self->{onerror} = $onerror; |
| 304 |
|
| 305 |
$self->{must_level} = 'm'; |
| 306 |
$self->{fact_level} = 'f'; |
| 307 |
$self->{should_level} = 's'; |
| 308 |
$self->{good_level} = 'w'; |
| 309 |
|
| 310 |
$self->{pluses} = {}; |
| 311 |
$self->{minuses} = {}; |
| 312 |
$self->{id} = {}; |
| 313 |
$self->{term} = {}; |
| 314 |
$self->{usemap} = []; |
| 315 |
$self->{contextmenu} = []; |
| 316 |
$self->{map} = {}; |
| 317 |
$self->{menu} = {}; |
| 318 |
$self->{has_link_type} = {}; |
| 319 |
#$self->{has_uri_attr}; |
| 320 |
#$self->{has_hyperlink_element}; |
| 321 |
#$self->{has_charset}; |
| 322 |
$self->{return} = { |
| 323 |
class => {}, |
| 324 |
id => $self->{id}, table => [], term => $self->{term}, |
| 325 |
}; |
| 326 |
|
| 327 |
my @todo = ({type => 'element', node => $el}); |
| 328 |
while (@todo) { |
| 329 |
my $todo = shift @todo; |
| 330 |
if ($todo->{type} eq 'element') { |
| 331 |
my $prefix = $todo->{node}->prefix; |
| 332 |
if (defined $prefix and $prefix eq 'xmlns') { |
| 333 |
$self->{onerror} |
| 334 |
->(node => $todo->{node}, level => 'NC', |
| 335 |
type => 'Reserved Prefixes and Namespace Names:<xmlns:>'); |
| 336 |
} |
| 337 |
my $nsuri = $todo->{node}->namespace_uri; |
| 338 |
$nsuri = '' unless defined $nsuri; |
| 339 |
unless ($Namespace->{$nsuri}->{loaded}) { |
| 340 |
if ($Namespace->{$nsuri}->{module}) { |
| 341 |
eval qq{ require $Namespace->{$nsuri}->{module} } or die $@; |
| 342 |
} else { |
| 343 |
$Namespace->{$nsuri}->{loaded} = 1; |
| 344 |
} |
| 345 |
} |
| 346 |
my $ln = $todo->{node}->manakai_local_name; |
| 347 |
my $eldef = $Element->{$nsuri}->{$ln} || |
| 348 |
$Element->{$nsuri}->{''} || |
| 349 |
$ElementDefault; |
| 350 |
$eldef->{attrs_checker}->($self, $todo); |
| 351 |
my ($new_todos) = $eldef->{checker}->($self, $todo); |
| 352 |
unshift @todo, @$new_todos; |
| 353 |
} elsif ($todo->{type} eq 'element-attributes') { |
| 354 |
my $prefix = $todo->{node}->prefix; |
| 355 |
if (defined $prefix and $prefix eq 'xmlns') { |
| 356 |
$self->{onerror} |
| 357 |
->(node => $todo->{node}, level => 'NC', |
| 358 |
type => 'Reserved Prefixes and Namespace Names:<xmlns:>'); |
| 359 |
} |
| 360 |
my $nsuri = $todo->{node}->namespace_uri; |
| 361 |
$nsuri = '' unless defined $nsuri; |
| 362 |
unless ($Namespace->{$nsuri}->{loaded}) { |
| 363 |
if ($Namespace->{$nsuri}->{module}) { |
| 364 |
eval qq{ require $Namespace->{$nsuri}->{module} } or die $@; |
| 365 |
} else { |
| 366 |
$Namespace->{$nsuri}->{loaded} = 1; |
| 367 |
} |
| 368 |
} |
| 369 |
my $ln = $todo->{node}->manakai_local_name; |
| 370 |
my $eldef = $Element->{$nsuri}->{$ln} || |
| 371 |
$Element->{$nsuri}->{''} || |
| 372 |
$ElementDefault; |
| 373 |
$eldef->{attrs_checker}->($self, $todo); |
| 374 |
} elsif ($todo->{type} eq 'plus' or $todo->{type} eq 'minus') { |
| 375 |
$self->_remove_minuses ($todo); |
| 376 |
} elsif ($todo->{type} eq 'code') { |
| 377 |
$todo->{code}->(); |
| 378 |
} else { |
| 379 |
die "$0: Internal error: Unsupported checking action type |$todo->{type}|"; |
| 380 |
} |
| 381 |
} |
| 382 |
|
| 383 |
for (@{$self->{usemap}}) { |
| 384 |
unless ($self->{map}->{$_->[0]}) { |
| 385 |
$self->{onerror}->(node => $_->[1], type => 'no referenced map'); |
| 386 |
} |
| 387 |
} |
| 388 |
|
| 389 |
for (@{$self->{contextmenu}}) { |
| 390 |
unless ($self->{menu}->{$_->[0]}) { |
| 391 |
$self->{onerror}->(node => $_->[1], type => 'no referenced menu'); |
| 392 |
} |
| 393 |
} |
| 394 |
|
| 395 |
delete $self->{pluses}; |
| 396 |
delete $self->{minuses}; |
| 397 |
delete $self->{onerror}; |
| 398 |
delete $self->{id}; |
| 399 |
delete $self->{usemap}; |
| 400 |
delete $self->{map}; |
| 401 |
return $self->{return}; |
| 402 |
} # check_element |
| 403 |
|
| 404 |
sub _add_minuses ($@) { |
| 405 |
my $self = shift; |
| 406 |
my $r = {}; |
| 407 |
for my $list (@_) { |
| 408 |
for my $ns (keys %$list) { |
| 409 |
for my $ln (keys %{$list->{$ns}}) { |
| 410 |
unless ($self->{minuses}->{$ns}->{$ln}) { |
| 411 |
$self->{minuses}->{$ns}->{$ln} = 1; |
| 412 |
$r->{$ns}->{$ln} = 1; |
| 413 |
} |
| 414 |
} |
| 415 |
} |
| 416 |
} |
| 417 |
return {type => 'plus', list => $r}; |
| 418 |
} # _add_minuses |
| 419 |
|
| 420 |
sub _add_pluses ($@) { |
| 421 |
my $self = shift; |
| 422 |
my $r = {}; |
| 423 |
for my $list (@_) { |
| 424 |
for my $ns (keys %$list) { |
| 425 |
for my $ln (keys %{$list->{$ns}}) { |
| 426 |
unless ($self->{pluses}->{$ns}->{$ln}) { |
| 427 |
$self->{pluses}->{$ns}->{$ln} = 1; |
| 428 |
$r->{$ns}->{$ln} = 1; |
| 429 |
} |
| 430 |
} |
| 431 |
} |
| 432 |
} |
| 433 |
return {type => 'minus', list => $r}; |
| 434 |
} # _add_pluses |
| 435 |
|
| 436 |
sub _remove_minuses ($$) { |
| 437 |
my ($self, $todo) = @_; |
| 438 |
if ($todo->{type} eq 'minus') { |
| 439 |
for my $ns (keys %{$todo->{list}}) { |
| 440 |
for my $ln (keys %{$todo->{list}->{$ns}}) { |
| 441 |
delete $self->{pluses}->{$ns}->{$ln} if $todo->{list}->{$ns}->{$ln}; |
| 442 |
} |
| 443 |
} |
| 444 |
} elsif ($todo->{type} eq 'plus') { |
| 445 |
for my $ns (keys %{$todo->{list}}) { |
| 446 |
for my $ln (keys %{$todo->{list}->{$ns}}) { |
| 447 |
delete $self->{minuses}->{$ns}->{$ln} if $todo->{list}->{$ns}->{$ln}; |
| 448 |
} |
| 449 |
} |
| 450 |
} else { |
| 451 |
die "$0: Unknown +- type: $todo->{type}"; |
| 452 |
} |
| 453 |
1; |
| 454 |
} # _remove_minuses |
| 455 |
|
| 456 |
## NOTE: Priority for "minuses" and "pluses" are currently left |
| 457 |
## undefined and implemented inconsistently; it is not a problem for |
| 458 |
## now, since no element belongs to both lists. |
| 459 |
|
| 460 |
sub _check_get_children ($$$) { |
| 461 |
my ($self, $node, $parent_todo) = @_; |
| 462 |
my $new_todos = []; |
| 463 |
my $sib = []; |
| 464 |
TP: { |
| 465 |
my $node_ns = $node->namespace_uri; |
| 466 |
$node_ns = '' unless defined $node_ns; |
| 467 |
my $node_ln = $node->manakai_local_name; |
| 468 |
if ($HTMLTransparentElements->{$node_ns}->{$node_ln}) { |
| 469 |
if ($node_ns eq $HTML_NS and $node_ln eq 'noscript') { |
| 470 |
if ($parent_todo->{flag}->{in_head}) { |
| 471 |
# |
| 472 |
} else { |
| 473 |
my $end = $self->_add_minuses ({$HTML_NS, {noscript => 1}}); |
| 474 |
push @$sib, $end; |
| 475 |
|
| 476 |
unshift @$sib, @{$node->child_nodes}; |
| 477 |
push @$new_todos, {type => 'element-attributes', node => $node}; |
| 478 |
last TP; |
| 479 |
} |
| 480 |
} else { |
| 481 |
unshift @$sib, @{$node->child_nodes}; |
| 482 |
push @$new_todos, {type => 'element-attributes', node => $node}; |
| 483 |
last TP; |
| 484 |
} |
| 485 |
} |
| 486 |
if ($node_ns eq $HTML_NS and ($node_ln eq 'video' or $node_ln eq 'audio')) { |
| 487 |
if ($node->has_attribute_ns (undef, 'src')) { |
| 488 |
unshift @$sib, @{$node->child_nodes}; |
| 489 |
push @$new_todos, {type => 'element-attributes', node => $node}; |
| 490 |
last TP; |
| 491 |
} else { |
| 492 |
my @cn = @{$node->child_nodes}; |
| 493 |
CN: while (@cn) { |
| 494 |
my $cn = shift @cn; |
| 495 |
my $cnt = $cn->node_type; |
| 496 |
if ($cnt == 1) { |
| 497 |
my $cn_nsuri = $cn->namespace_uri; |
| 498 |
$cn_nsuri = '' unless defined $cn_nsuri; |
| 499 |
if ($cn_nsuri eq $HTML_NS and $cn->manakai_local_name eq 'source') { |
| 500 |
# |
| 501 |
} else { |
| 502 |
last CN; |
| 503 |
} |
| 504 |
} elsif ($cnt == 3 or $cnt == 4) { |
| 505 |
if ($cn->data =~ /[^\x09-\x0D\x20]/) { |
| 506 |
last CN; |
| 507 |
} |
| 508 |
} |
| 509 |
} # CN |
| 510 |
unshift @$sib, @cn; |
| 511 |
} |
| 512 |
} |
| 513 |
push @$new_todos, {type => 'element', node => $node}; |
| 514 |
} # TP |
| 515 |
|
| 516 |
for my $new_todo (@$new_todos) { |
| 517 |
$new_todo->{flag} = {%{$parent_todo->{flag} or {}}}; |
| 518 |
} |
| 519 |
|
| 520 |
return ($sib, $new_todos); |
| 521 |
} # _check_get_children |
| 522 |
|
| 523 |
=head1 LICENSE |
| 524 |
|
| 525 |
Copyright 2007 Wakaba <w@suika.fam.cx> |
| 526 |
|
| 527 |
This library is free software; you can redistribute it |
| 528 |
and/or modify it under the same terms as Perl itself. |
| 529 |
|
| 530 |
=cut |
| 531 |
|
| 532 |
1; |
| 533 |
# $Date: 2007/10/14 09:21:46 $ |
Parent Directory
| 
Revision Log